Thursday, April 11, 2013

Navigating the complexities of authorship: Part 1 –inclusion

One of the highlights of grad school is publishing your very first papers in peer-reviewed journals. I can still remember the feeling of seeing my first paper appear in print (yes on paper and not a pdf). But what this novice scientist should not be fretting over is which colleagues should be included as authors and whether they are breaking any norms. The two things that should be avoided are including as authors, those that did not substantially contribute to the work, and excluding those that deserve authorship. There have been controversial instances where breaking these authorship rules caused uncomfortable situations. None of us would want someone writing a letter to a journal arguing that they deserved authorship. Nor is it comfortable to see someone squirming out of authorship, arguing they had minimal involvement when an accusation of fraud has been levelled against a paper. How to determine who should be an author can be difficult.

Even though I spell out my own rules below, it is important to be flexible and to understand that different types of papers and differing situations can have an impact on this decision. That said, you do not want to be arbitrary in this decision. For example, if two people contribute similar amounts to a paper, you do not want to include only one because you personally dislike the other. You should have a benchmark for inclusion that can be defended. The cartoon above highlights the complexity and arbitrariness of authorship –and the perception that there are many instances of less than meritorious inclusion.

Journals do have their own guidelines, and many now require statements about contributions, but even these can be vague, still making it difficult to assess how much individuals actually contributed. When I discuss issues of authorship with my own students, I usually reiterate the criteria from Weltzin et al. (2006). I use four criteria to evaluate contribution:
1)   Origination of the idea for the study. This would include the motivation for the study, developing the hypotheses and coming up with a plan to test hypotheses.
2)   Running the experiment or data collection. This is where the blood, sweat and tears come in.
3)   Analyzing the data. Basically moving from a database to results, including deciding on the best analyses, programming (or using software) and dealing with inevitable complexities, issues and problems.
4)   Writing the paper. Putting everything together can sometimes be the most difficult and external motivation can be important.

My basic requirements for authorship are that one of these steps was not possible without a key person, or else there was a person who significantly contributed to more than one of these. Such requirements mean that undergraduates assisting with data collection do not meet the threshold for authorship. Obviously these are idealized and different types of studies (e.g., theory or methodological papers) do not necessarily have all these activities. Regardless, authors must have contributed in a meaningful way to the production of this research and should be able to defend it. All authors need to sign off on the final product.

While this system is idealized, there are still complexities making authorship decisions difficult or uncomfortable. Here are three obvious ones –but there are others.

Data sharing
Large, synthetic analyses require multiple datasets and some authors are loath to share their hard work without credit. This is understandable, as a particular dataset could be the product of years of work. But when is inclusion for authorship appropriate? It is certainly appropriate to offer authorship if the questions being asked in the synthesis overlap strongly with planned analyses for the dataset. Both the data owner and the synthesis architect have a mutual interest in fostering collaboration. In this case every effort should be made to include the data owner in the analyses and writing of the manuscript.

When is it not appropriate to include data owners as authors? First and foremost, if the data is publically available, then it is there for further independent investigation. No one would offer authorship to each originator of a gene sequence in Genbank. Secondly, if it is a dataset that has already been used in many publications and has fulfilled its intended goals, then it should be made available without authorship strings. I’ve personally seen scientists reserve the right of authorship for the use of datasets that are both publically available and have satisfied the intended purpose long ago.

The basic rule of thumb should be that if the dataset is recent and still being analyzed, and if the owner has an interest in examining similar questions, then authorship should be offered –with the caveat that additional work is required, beyond simply supplying the data.

Idea ontogeny
I thought about labeling this section ‘idea stealing’ but thought that wasn’t quite right. An idea is a complex entity. It lives, dies and morphs. It is fully conceivable to listen to a news story about agricultural subsidies, which somehow spurs an idea about ecosystem dynamics. We all have conversations with colleagues and go to talks, and these interactions can morph into new scientific ideas, even subconsciously. We need to be careful and acknowledge how much an idea came from a direct conservation with another scientist. Obviously if a scientist says “you should do this experiment…”,  then you need to acknowledge them and perhaps turn your idea into a collaboration.

Now here is the tricky one. Often people are authors because they control the purse strings. Yes, a PI has done an excellent job of securing funding, and should be acknowledged for this. If the study is a part of a funded project, where the PI developed the original idea, then the PI fully deserves to be included. However, if the specific study is independent from the funded project in terms of ideas and work plan, but uses funding from this project, then this contribution belongs in the acknowledgements and does not deserve authorship. There are cases where the PI of an extremely large lab gets dozens of papers a year, always appearing last in the list of authors (see part 2 on author order -forthcoming), and it is legitimate to view their contributions skeptically. Their relationship to many of the papers is likely financial and they probably couldn’t defend the science. I had a non-ecologist colleague ask me if it was still the case that graduate students in ecology produce papers without their advisors, to which I said yes (Caroline has several papers without me as an author).

Clearly there are cultural differences among sub disciplines. However, I do feel that authorship norms need to be robust and enforced. Cheaters (those gratuitously appearing on numerous papers –see part 3 on assigning credit; also forthcoming) reap the rewards and benefits of authorship, with little cost. It is disingenuous to list authors that have not have a substantial input into the publication, and the lead author is responsible for the accuracy of authorship. The easiest way to ensure that authors are really authors is to make an effort to include them in various aspects of the paper. For example, give them every opportunity to provide feedback –send them the first results and early drafts, have Skype for phone meetings with them to get their input and incorporate that input. Ultimately, we all should walk away from a collaboration feeling like we have contributed and made the paper better, and we should be proud to talk about it to other colleagues.

Many of these ideas were directly informed by this great paper by Weltzin and colleagues (2006):

Weltzin, J. F., Belote, R. T., Williams, L. T., Keller, J. K. & Engel, E. C. (2006) Authorship in ecology: attribution, accountability, and responsibility. Frontiers in Ecology and the Environment, 4, 435-441. 


Jeremy Fox said...

Nice post Marc. Put up my own thoughts on this (broadly in line with yours, but with some differences) here:

Sparked a good discussion. Probably the single key thing everyone agreed on is for all authors (or candidate authors) to discuss authorship well in advance, ideally before even embarking on a project.

Where there was some disagreement was over whether paying for the work and providing general supervision itself merits authorship, and on precise thresholds for inclusion. I have the sense that we're all becoming more generous about co-authorship, to the point where I probably now set a higher-than-average bar for authorship (thereby likely denying myself at least a few co-authored papers that many other supervisors in my position would have had!)

Marc Cadotte said...

Thanks Jeremy. Yes, the cost of not being included gratuitously on papers is that others are. The paper inflation (increasing average number of papers per year) wins in consideration for awards, grants, etc. Unfortunately there isn't a direct benefit from following rigorous authorship rules -except for knowing you did the right thing.